Neko 1.99.9
A portable framework for high-order spectral element flow simulations
Loading...
Searching...
No Matches
Static Public Member Functions | List of all members
cdtp_mfma_dispatch< T, LX, NWF > Struct Template Reference

#include <cdtp_kernel.h>

Static Public Member Functions

static __device__ void run (T *, const T *, const T *, const T *, const T *, const T *, const T *, const T *, const T *, const int)
 

Detailed Description

template<typename T, const int LX, const int NWF>
struct cdtp_mfma_dispatch< T, LX, NWF >

Device kernel for D^T x on the AMD matrix cores

The only one of these whose contractions come after its pointwise work rather than before it, and the only one that accumulates: the three weighted fields are formed first and then contracted into a single output. That is axhelm's second half rather than its first.

Because the operator is handed dxt, dyt and dzt already transposed, the contraction index pattern is the ordinary forward one, so mfma_contract_sel is used with TRANSPOSE = false against the staged transposes rather than with TRANSPOSE = true against the forward matrices. The output is zeroed with the staging and every axis accumulates into it, matching the divergence half of ax_helm_mfma_elem().

Definition at line 325 of file cdtp_kernel.h.

Member Function Documentation

◆ run()

template<typename T , const int LX, const int NWF>
static __device__ void cdtp_mfma_dispatch< T, LX, NWF >::run ( T *  ,
const T *  ,
const T *  ,
const T *  ,
const T *  ,
const T *  ,
const T *  ,
const T *  ,
const T *  ,
const int   
)
inlinestatic

Definition at line 326 of file cdtp_kernel.h.

Here is the caller graph for this function:
The documentation for this struct was generated from the following file: